IU researchers license AI tool to diagnose Alzheimer’s
Subscriber Benefit
As a subscriber you can listen to articles at work, in the car, or while you work out. Subscribe Now
Dr. Malaz Boustani, an esteemed brain doctor at the Indiana University School of Medicine, is haunted by the fact that he didn’t identify his own father’s Alzheimer’s disease. “I think I’m a great brain doctor—and I missed it,” he says.
Boustani believes his father was a victim of the system, which he says currently misses about 90% of people in the U.S. who have mild cognitive impairment or dementia. Boustani and another IU researcher are confident they’ve built the solution: an artificial intelligence tool with 80% accuracy in detecting patients with undiagnosed Alzheimer’s disease and related forms of dementia.
Recently licensed by tech startup DigiCARE Realized, the technology is now taking aim at commercialization.
Boustani, a research scientist with the Regenstrief Institute and IUSM Richard M. Fairbanks professor of aging research, says current diagnostic methods for mild cognitive impairment and dementia lack scalability—a major reason why so many cases are missed. Detecting biological markers require a spinal tap, PET scans to identify amyloid plaque in the brain are very expensive, and other less invasive assessments are too time consuming to be widely deployed, say brain experts.
With scalability as a major goal, Boustani and IUPUI Associate Professor of Electrical and Computer Engineering Dr. Zina Ben Miled have created an AI tool that combs through electronic health records, or EHRs, to detect early Alzheimer’s and related dementia.
“[The tool] uses data that’s already been collected and stored in EHRs for millions and millions of Americans; the data has already been collected, so it doesn’t require the investment of any new time,” says Boustani. “We use a machine learning algorithm that takes passive data, processes it and spits out an output that classifies your risk of having Alzheimer’s and other related dementia.”
Within EHRs, the tool is crunching two types of data; the first is “structured” data, such as billing codes commonly used in the healthcare system, and medications.
“But the big, most important type of data our algorithm goes through is what we call ‘unstructured data,’” says Boustani. “These are the notes—almost like texts—the doctor, social worker or nurse makes when writing something in the EHR, such as the reports from an EKG. It’s like the written notes [in EHRs].”
The IU Innovation and Commercialization Office says the tool had an 80% accuracy rate in three-year prediction horizons when tested at about 100 hospitals throughout Indiana. Boustani says because “the nerdy scientists in us are skeptical, we want validation after validation,” so the researchers also tested the tool in Pennsylvania, where it showed equal accuracy after combing through EHRs.
DigiCARE Realized, a tech startup on the prowl for “pioneers in brain health” took notice and recently licensed the technology.
“It was very exciting to see it demonstrate that performance accuracy in mild cognitive impairment, even a stage earlier than Alzheimer’s disease on the continuum,” says DigiCARE co-founder and CEO Brittany Cassin. “We recognize the power of AI—which I think, oftentimes, is such a buzz word—but the problem is, cognitive care is not standardized. Being able to deliver cognitive care in an equitable, accessible and patient-centered way is the power of what the AI approach is bringing.”
Boustani and Miled fill out the DigiCARE founding team as chief health officer and chief information officer, respectively. The team envisions the tool being used within health systems as a routine cognitive care method.
“The next step is working with health systems that are excited about the power of AI and looking at applying this proactive approach in their health systems,” says Cassin. “We’re also planning to start fundraising for our pre-seed round and to take us through our first phase of commercialization.”
DigiCARE says the emergence of new Alzheimer’s treatments makes early detection even more critical. In January, the FDA approved LEQEMBI, a medication to treat Alzheimer’s disease in its early stages. The team believes identifying Alzheimer’s in its earliest stages will also help the effectiveness of other emerging treatments that focus on symptoms and later stages.
“All of the work I’ve been doing at IUSM for 20-plus years has been starting with discovery and ending on the bookshelf,” says Boustani. “But this commercialization will move it from [a journal article] on the shelf to an actual real-world solution that will change lives on a scale.”
While DigiCARE Realized is legally based in New Jersey, Cassin says it’s a “virtual first” company that’s “very much connected to Indiana.”